Study On The Application Of Combined Classifiers In The Prediction Of Kidney Disease Risk

In the new era of socialism with Chinese characteristics,the key to improving peopleundefineds living standards,ensuring the convenience of medical treatment and increasing the life expectancy lies not only in the strength of the medical team,but also in the development and innovation of medical technology.It is of great practical significance to study a reliable,practical and good kidney disease risk prediction system for early prevention and control of diseases.Risk prediction is a decision-making process.On the basis of the poor prediction performance of single classifiers,the method of combined classifiers is selected for risk prediction.Combined classifiers can effectively improve the application generalization ability of risk prediction system,and have been widely concerned by many scholars in recent years.After collecting nearly 12806 original data of kidney disease,a total of 10000 valid data were selected after data preprocessing and data specification,and the database of kidney disease risk prediction was established.After modeling four single classifiers algorithm,such as SVM,BP neural network,C4.5 and Bayes,Adaboost,Bagging and Stacking is used to model four single classifiers algorithm.The algorithm of Stacking and other combined classifiers is used to further classify and predict the data.It is proved that the performance of combined classifiers is better than that of single classifiers on the basis of kidney disease data.The main work follows:1.The background,practical value and application prospect of kidney disease risk prediction are introduced,and the research status and main research contents of kidney disease risk prediction at home and abroad are studied.2.The main methods of machine learning are studied,including support vector machine classification algorithm,C4.5 decision tree classification algorithm,naive Bayesian classification algorithm,BP neural network classification algorithm,Adaboost algorithm,Bagging algorithm,Stacking algorithm.The concept of algorithm,algorithm principle,algorithm steps and typical applications are described respectively.3.Clean the collected kidney disease data,select the features,reduce the dimension of the features,and standardize the unit and value range of the kidney disease examination value.The construction of kidney disease risk prediction system has the basic operation function and model establishment and risk prediction function of the data.4.Through the analysis of the results of the risk prediction system,the existing theory of kidney disease risk prediction is verified,and the new conclusion of kidney disease risk prediction is explored and found,which guides the further experimental research and clinical diagnosis and treatment.